Lean premixed combustion systems have been deployed on land based gas turbine engines to reduce emissions, such as NOx and CO. These systems have been successful and, in some cases, produce emission levels that are at the lower limits of measurement capabilities, approximately 1 to 3 parts per million (ppm) of NOx and CO. Although these systems are a great benefit from a standpoint of emission production, the operational envelope of the systems is substantially reduced when compared to more conventional combustion systems. As a consequence, the control of fuel conditions, distribution and injection into the combustion zones has become a critical operating parameter and requires frequent adjustment, when ambient atmospheric conditions, such as temperature, humidity and pressure, change. The re-adjustment of the combustion fuel conditions, distribution and injection is termed tuning.
Controlled operation of a combustion system generally employs a manual setting of the operational control settings of a combustor to yield an average operational condition. These settings may be input through a controller, which as used herein shall mean any device used to control the operation of a system. Examples include a Distributed Control System (DCS), a gas turbine controller, a programmable logical controller (PLC), a stand-alone computer with communication to another controller and/or directly to a system.
These settings are satisfactory at the time of the setup, but conditions may change when tuning issues arise and cause an unacceptable operation in a matter of hours or days. Tuning issues are any situation whereby any operational parameters of a system are in excess of acceptable limits. Examples include emissions excursion outside of allowable limits, combustor dynamics excursion outside of allowable limits, or any other tuning event requiring adjustment of a turbine's operational control elements. Other approaches use a formula to predict emissions based on a gas turbine's operating settings and select a set point for fuel distribution and/or overall machine fuel/air ratio, without modifying other control elements, such as fuel gas temperature. These approaches do not allow for timely variation, do not take advantage of actual dynamics and emission data or do not modify fuel distribution, fuel temperature and/or other turbine operating parameters.
Another variable that impacts the lean premixed combustion system is fuel composition. Sufficient variation in fuel composition will cause a change in the heat release of the lean premixed combustion system. Such change may lead to emissions excursions, unstable combustion processes, or even blow out of the combustion system.
Mis-operation of the combustion system manifests itself in augmented pressure pulsations or an increase in combustion dynamics (hereinafter, combustion dynamics may be indicated by the symbol “HP”). Pulsations can have sufficient force to destroy the combustion system and dramatically reduce the life of combustion hardware. Additionally, improper tuning of the combustion system can lead to emission excursions and violate emission permits. Therefore, a means to maintain the stability of the lean premixed combustion systems, on a regular or periodic basis, within the proper operating envelope, is of great value and interest to the industry. Additionally, a system that operates by utilizing near real-time data, taken from the turbine sensors, would have significant value to coordinate modulation of operational control elements such as fuel distribution, fuel gas inlet temperature and/or overall machine fuel/air ratio.
While real-time tuning of a combustion system can provide tremendous operational flexibility and protection for turbine hardware, a combustion system may concurrently experience a number of different operational issues. For example, most turbine operators of lean premixed combustion systems are concerned with exhaust emissions (NOx and CO) as well as combustor dynamics. It is not uncommon for both high NOx emissions and high combustor dynamics to coexist on a turbine. Additionally, tuning in response to one concern can make other constraints worse, for example tuning for low NOx can make combustor dynamics worse, tuning for high CO can make NOx worse, etc. It would be beneficial to provide a system whereby an algorithm is used to compare the current status of all tuning concerns, rank each concern in order of importance, determine the operational concern of most interest, and subsequently commence automated tuning to remediate this dominant operational concern.